I have an application where an HX711 ADC is used with 4 load cells. (in wheatstone bridge configuration, rated at 50kg each).

The thing is that I am trying to sense 30~ish gram changes with it accurately, while the cells will be loaded with 10 to 40 kilograms.

I have built up the circuitry and placements properly, calibrated and tested etc. but I have 2 questions that I couldn't come up with a solution.

1st question: When powering arduino from USB cable the readings are much more stable (they are fluctuating like +-5 grams only) but when I connect a 12V power source (with 5A output) to the arduino, the readings start to fluctuate much more. (+-30 grams or so). The HX711 is powered from arduino 5v out pin but even though i tried to power it from a 5v adapter the fluctuations seemed to happen.

I should also note that the HX711 is connected to the analog pins A0 and A1, and there are 8 other sensors connected to the pwm pins of the arduino.

Now any ideas why does that happen? Any way to prevent this fluctuation?

2nd question: If I calibrate with a load close to the max. weight (as suggested in many papers) the small changes seems to be very unreliable. If I calibrate it with a small weight (like around 200g) I can sense small weights properly but this time when a 2 kg weight is put it is read like 2.3kg or something. Is this normal?

I know that my load cells are not the proper ones for the application but I would like to push them as much as i can.

Thanks for your time

  • Try a few different adapters. Some generate very noisy power. Secondly, you could add multiple calibration points. Cause the output, as you noticed, is not linear.
    – Gerben
    Nov 10, 2015 at 14:24
  • I will try different sources and return an answer asap, thank you. Nov 10, 2015 at 15:13

1 Answer 1


1) Put your power source on an oscilloscope and check it. I assume what's happening is you don't have a clean sine, and the power fluctuates, whereas power from a PC USB port is usually very well regulated. This may be causing issues in your sensor due to under/over voltage at a minuscule level.

2) Because you're pushing your sensors to the edge of their capabilities, you're going to have to work around them, so the normal scenario's don't really offer much help, as they are usually based on working within the middle ground of capabilities.

You said that factoring based on low load vs high load created unreliable variances in your measurements, i would personally suggest calibrating not for max load, nor min load, but 2/3 of max, or 1/3 max, and see if that stabilizes you, or at least gets you somewhere to the middle of the road where your differences are nominal, or something you can live with.

  • Thank you for your comments and suggestions! I did not have an oscilloscope at hand, i had to assume the power source was working clean as it was somewhat commercial design but I guess i'll have to check it, it really seems to be the problem. Nov 10, 2015 at 13:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.